Work / 02
Google Ads · Measurement
Half the clicks, twice the right ones
Qualified enquiries from senior buyers of business services, through Google Search, on a fixed monthly budget. The account had been run hands-off and report-driven. It was taken in-house, read for two weeks, then rebuilt.
The five steps
Step 1AuditTwo weeks reading the account before changing anything.
Step 2FindFour problems, each proven with a report.
Step 3FixSix changes, in order of impact.
Step 4MeasureWeekly reconciliation of leads against the sales tracker.
Step 5ResultSpend reaching real buyers up. Counting honest.
The four problems
Each one was shown with the report that proved it before anything changed.
Most spendLoose keywords. Broad match. Search terms full of job seekers, other company names and one-word searches.
About 25%Hidden placements. Search partners and display switched on inside search campaigns. Click rates no buyer produces.
9 in 10False conversions. One event fired on every website form, job applications included. Bidding learned from it.
About 50%Lost to rank, no rival. Half the impressions lost to rank while Auction Insights showed no competitor. A quality problem.
What changed
| Area | Before | After |
|---|---|---|
| Keyword matching | Mostly broad match | Phrase and exact only. Zero-volume keywords out. New reach keywords from Keyword Planner, so coverage went up while junk went down. |
| Blocked searches | A short list | About 180 new blocked terms: job words, localities, competitor names, one-word terms. |
| Placements | Partners and display on | Both off. Spend stays on Google Search results. |
| Conversion goal | One site-wide event | One goal per campaign that counts only that campaign's own form. |
| Lead tracking | Form entries, no source | Click ID and campaign saved with every lead. Weekly check against the sales tracker. |
| Ad groups and copy | Generic | Rewritten around real search terms. Brand, generic and competitor searches split into their own campaigns. |
Results
Junk traffic out, so clicks fell and the click price rose. That was the plan. The money now reaches people who were actually searching for the service.
28% to 50%of spend on right-intent searches, campaign A
About 3xright-intent share of spend, campaign B
Halvedclicks, by design. The removed clicks were the wrong people.
Matchreported leads and real leads, every week since the counting fix
| Question | Before | After |
|---|---|---|
| Where does the money go? | Broad searches, partner sites, display | Google Search results only, phrase and exact match |
| What counts as a lead? | Any form on the site | That campaign's own form, checked against sales weekly |
| Why are we losing impressions? | Assumed competition | Proven to be relevance. Fixed with copy and pages, not bids |
| Who runs it? | Hands-off, report-driven | Weekly, with the search term report open |
Every week since
Search termsBlock and addNew junk terms blocked, new reach terms added. The list grows every week.
PlacementsStays at zeroSpend on search partners and display, checked so it never creeps back.
LeadsSheet against salesEvery reported lead found in the lead sheet with its click ID and campaign.
Rank and qualityBy ad groupLost-to-rank share and quality score. Copy or page fixed where either slips.
What this taught me
- Cheap clicks are a warning sign, not a win.
- Fix the counting before the bidding. Smart bidding on a bad signal spends confidently in the wrong direction.
- Audit before you touch. Two weeks of reading saved months of guessing.
Next on this account
- Send sales-accepted lead status back to Google Ads as offline conversions.
- Rebuild the landing pages: form near the top, fewer fields, faster on mobile.
- Test one campaign on a lead-quality bid target once 30 days of clean data exist.
Happy to open the search term report, the blocked list, the conversion goals and the lead sheet with click IDs on a call.